MichalCervenansky/Automatic-image-annotation-for-microstock-sites ? reverse-engineered prompt

Build me a Python tool that can automatically tag photos for microstock sites like Shutterstock or Fotolia.

I want to be able to run it on one image or a whole folder of images, and also pick files from a popup if I don’t pass any arguments. It should read any existing IPTC keywords and caption text from the image, detect objects in the photo, classify the image itself, and then combine all of that into a good set of stock style keywords. If possible, use the detected and existing keywords as seed words to improve the final suggestions.

Please save the finished keywords back into the image’s IPTC metadata so the file is ready to upload. I’d like the behavior to be easy to tweak from a config file, especially which annotation methods are used. If you need to look up current docs for TensorFlow or any model downloads, go ahead.